As mentioned, Nsmce2SD mice bear some of features typically associated with genomic instability, at least at the cellular level. Considering that the observed phenotypes overlapped with what described for Blm mutants, we reasoned that a classical genetic approach might help elucidate the functional role of the SMC5/6 complex in dissolution.
In order to test such hypothesis, we decided to take advantage of mouse models for resolution described in the literature.
A first available candidate was the Mus81 knock-out mouse reported in McPherson, Lemmers et al. 2004. We rederived the strain on our mixed-background line, verifying at first the actual deficiency of the protein by Western blot. (figure 23 - A)
"
Figure 23: Mus81KO animals show a genomic instability signature. (A) We could confirm the actual deletion of MUS81in rederived animals by Western blot on testis protein samples. (B) A reported feature of Mus81KO cells is a mild sensitivity to MMC. T-lymphocytes from Mus81KO animals exposed to the drug accumulate chromosomal aberrancies as the ones highlighted by white arrows (adapted from McPherson, Lemmers et al. 2004)
Interestingly, the Mus81KO strain we used was reported to show haplosufficient tumour suppression, which reflected in the survival rates of animals (figure 24-A). This striking result is in contrast with what observed in an alternative Mus81-null strain (Dendouga, Gao et al. 2005), which instead lacked any major phenotype (figure 24-B), and such
"60
A B
Results - part I important contradiction warned us on the possibility of a background-related effect for the mutation.
Despite the remarkable survival discrepancy, both models concurred on a mild sensitivity to agents such as MMC, UV irradiation and MMS, as in figure 25.
!
Figure 24: discrepancies between Mus81KO models available. The two Mus81-null strains reported in the literature show a strikingly contradictory difference in terms of survival.(A) KM survival curve for the Mus81KO model used in our study.(B) KM survival curve from the alternative Mus81 model (McPherson et al. 2004;
Dendouga et al. 2005).
"
Figure 25: all Mus81KO strains show comparable sensitivity to DNA damaging agents. Both Mus81KO models respond consistently to DNA damaging drugs as highlighted by SCE experiments (A – McPherson et al. 2004) or metaphase spreads from B-cells (B – Dendouga et al. 2005)
A B
A B
A B
Results - part I Our first experimental approach aimed at verifying the outcome of impairing genetically both the resolution and the dissolution of DNA joint molecules, contemporarily.
Numerous reports, as previously mentioned, show the lethal outcome of this genetic combination in lower eukaryotes. According to what first observed in yeast (Kaliraman, Mullen et al. 2001), mus81 and mms4 are also synthetic lethal with null mutations in mus309, which encodes the orthologue of the Bloom syndrome helicase in Drosophila (Trowbridge, McKim et al. 2007).
A possibility we consequently envisaged from our experimental setup was synthetic sickness: the inability to cope with replication intermediates would lead to the accumulation of joint DNA species and result in increased genomic instability, especially in organs and tissues with a sustained cellular turnover and high rates of cellular replication.
We thus continued by crossing our Nsmce2SD animals onto the rederived Mus81KO strain, defining a genetic scenario by which both resolution and dissolution (considering the genetic equivalence Nsmce2 = Blm we postulated) would be affected.
The Nsmce2SD: Mus81KO mutations are not synergic in cells.
The dissection of the possible genetic interaction between Mus81 and Nsmce2 followed a
“simple to complex” approach. We started with experiments on MEFs derived from the Mus81KO strain, reasoning that the outcome of impairing resolution could be a feedback up-regulation of the dissolution pathway, following a trend commonly observed in biological systems (Kitano 2004).
As a proof of principle, we quantified the accumulation of NSMCE2 foci in Mus81KO cells, which could indicate an accumulation of joint DNA species demanding the activity of the SMC5/6 complex, as previously reported in our lab (unpublished data). We therefore performed NSMCE2 immunofluorescence stainings on Mus81KO and wild-type cells, after treating them with a sublethal dose of MMS. As shown in figure 26, NSMCE2 did aggregate into bigger and more intensely stained foci upon the treatment with MMS, but no marked difference between the two lines could be emphasized.
"62
A B
Results - part I
Figure 26: NSMCE2 foci formation is unaffected by MUS81 depletion. Primary fibroblasts from Mus81KO animals and wild-type controls were treated with MMS and and stained for NSMCE2/Mms21. Upon treatment, a similar accumulation of NSMCE2 foci was observed in both samples. Scale bar: 15µm.
These first negative observations don’t rule out a possible functional interaction between NSMCE2 and MUS81. For that reason we sought additional evidence from the Nsmce2SD : Mus81KO double mutant model we had just generated.
As a first step, we verified that the protein levels of NSMCE2 wouldn’t be affected by the deletion of Mus81 (figure 27).
"
Figure 27: NSMCE2 levels are unaltered after the deletion of Mus81 in cells. Cells from wild-type, Nsmce2SD : Mus81KO and relevant single-mutated controls were genotyped and blotted for NSMCE2. The levels of the protein were comparable to the wild-type controls in all genetic contexts.
A B
Results - part I We then proceeded along the line with what done for the Nsmce2SD model, and monitored the accumulation of micronuclei and polynucleation in MEFs obtained from Nsmce2SD : Mus81KO animals (and relevant controls).
As depicted in figure 28, double-mutant cells treated with increasing concentrations of MMS didn’t respond differently from controls. We noticed a trend in the accumulation of micronuclei that followed the increasing doses of MMS treatment (at least in the case of wild-type cells) but in the other genetic backgrounds, the situation wasn’t quite as defined.
Figure 28: Nsmce2SD:Mus81KO show similar micronuclei accumulation as their control counterparts.
When scoring for micronuclei accumulation in the double mutant cell subtype, we couldn’t highlight a negative synergy between mutated Nsmce2 and Mus81.
Mus81KO cells, expectedly, accumulated less micronuclei than their counterparts, due to their sensitivity to MMS, an alkylating agent reported to impair cell proliferation at the concentrations we employed. (McPherson, Lemmers et al. 2004)
Double mutants behaved accordingly to what observed in the single mutants, inheriting the characteristics of both Nsmce2SD and Mus81KO alleles. They showed a basal tendency to instability, though such tendency couldn’t be exacerbated by MMS treatment, probably because of the proliferation arrest driven by the absence of MUS81 in the presence of MMS, as previously reported (McPherson, Lemmers et al. 2004).
"64
B
Results - part I The quantification of polynucleated cells provided a similar outcome (figure 29): the basal levels of polynucleation in single mutants and double mutants were higher than wild-type controls, but the negative effect on proliferation impeded to highlight any incremental effect at increasing concentrations of the drug. The double-mutants, moreover, didn’t show a worsened response than the single mutants alone.
Figure 29: Nsmce2SD:Mus81KO cells tend to polynucleation spontaneously, similarly to single mutant controls. When scoring the polynucleation rates of single mutants and double mutant cells, we noticed a positive trend, but no direct correlation with the concentration of MMS used to induce damage.
To gain further insights into the DNA damage response of Nsmce2-Mus81 double mutant cells, we analyzed the accumulation of nuclear 53BP1 foci upon treatment with damaging agents (such as MMC and MMS).
We envisioned that the depletion of MUS81 in a NSMCE2SD background - hence, in a context of accumulation of joint DNA molecules - could result in the breakage of unprocessed recombination intermediates at cell division. Consequently, the rupture of intertwined DNA molecules would promote DNA repair and the accumulation of 53BP1 at the site of breaks. We thus interrogated MEFs from our murine model for their propensity to accumulate 53BP1 foci, comparing double Nsmce2SD - Mus81KO mutants with their controls.In order to increase the accumulation of replication intermediates, we induced the collapse of replication forks by treating cells with increasing concentrations of MMC and a sublethal dose of MMS. The results outlined in figure 30-A
A B
Results - part I highlight how double mutant cells show an accumulation of 53BP1 foci comparable to that of single Mus81KO and Nscmce2SD mutants, at different MMC concentrations.
Interestingly, the number of foci in both single and double mutant cells deviated from that of wt controls, suggesting, indirectly, the actual accumulation of broken DNA replication intermediates in these cells. As in the case of micro-nucleation and polynucleation though, no marked negative synergy could be detected.
Figure 30: Nsmce2SD-Mus81KO cells show an unaltered response to DNA damage upon the induction of replication stress. (A) 53BP1 foci were quantified by HT microscopy after the induction of replication stress (by means of MMC and MMS treatment) on double mutant cells and relevant controls. (B) Representative image of the induction of gH2AX and 53BP1 foci in cells upon treatment with MMC. Scale bar = 15 μm. (C) The accumulation of pan-nuclear gH2AX foci in cells (an established readout of replication stress) didn’t differ markedly among the different lines.
Finally, we checked for the accumulation of chromosomal abnormalities in double Nsmce2SD-Mus81KO cells and controls. Whereas we could detect the presence of aberrant chromosomal figures (such as radials, also commonly found in Blm cells, as in figure 31 - circled in red) we didn’t encounter a statistically significant difference between wt cells, Nsmce2SD, Mus81KO or the double Nsmce2SD- Mus81KO samples.
"66
A
B A
B
C
Results - part I
Figure 31: the Nsmce2SD-Mus81KO mutations do not result in aberrant metaphases in MEFs. Metaphase spreads were prepared from wild type, Nsmce2SD, Mus81KO and Nsmce2SD-Mus81KO cells, after treatment with either MMS or MMC, and chromosomal aberrancies were sought for (blue arrows). Though no striking effects were noticeable after MMS treatment, we highlighted the presence of radial chromosomes in MMC treated Nsmce2SD-Mus81KO cells.
Compensatory effects of Gen1-Slx1/4 in Nsmce2SD:Mus81KO cells?
It is possible that the lack of a clear phenotype in our cells could be due to compensatory pathways. The plausible functional redundancy of SLX1/SLX4 and GEN1 in mediating resolution could indeed counterbalance the absence of MUS81.
In order to shed a light on such possibility, we silenced, using RNAi, Slx4 and Gen1, with the aim of impairing resolution of joint DNA molecules significantly.
Our experimental approach is summarized in figure 32.
!
!
Results - part I
Figure 32: experimental strategy to impair resolution in Nsmce2SD-Mus81KO cells.(A) We used a lentiviral based platform to selectively silence Slx4 and Gen1 in double homozygous cells and controls with a commercially validated shRNA sequence. (B) After transduction and antibiotic selection, we measured the silencing of shRNA targets by RT-PCR, verifying their partial efficacy.
After transducing MEFs from our double mutant line and control strains with shRNAs against Gen1 and Slx4, we selected infected cells by adding puromycin to the culture medium.
Additionally, we immortalized them by infecting puromycin-selected pools with a retrovirally encoded SV40T121, in order to overcome the possible replicative arrest of MEFs after transductions.
We tested for the efficacy of the shRNA mediated silencing by measuring the RNA levels of Gen1 and Slx4 in our treated cells (figure 32-B), verifying how only a partial knock-down of nucleases could be attained. Despite the mild silencing, we established growth curves of shRNA treated MEFs, either primary or T121 immortalized, to verify if the partial silencing of resolution nucleases could nevertheless result in impaired growth in a Nsmce2SD-Mus81KO genetic background.
As in figure 33, all cell types responded aspecifically to the RNAi procedures, mostly by arresting proliferation (particularly in the case of Gen1 silencing). The same result was recapitulated when we tested both primary and immortalized MEFs for EdU incorporation (figure 34) Both the extensive manipulation and the partial functionality of the shRNAs employed (as well as plausible off-targeting effects of the shRNA employed) led to important effects on cell proliferation, jeopardizing the possibility of using this setup as an experimental platform.
"68
A B
Results - part I
"
Figure 33: the shRNA mediated silencing of endonucleases impairs cells proliferation aspecifically-
#1. Primary MEFs treated with shRNAs for Slx4, Gen1 (or an empty control) tend to reduce their proliferation rate and enter proliferative arrest. Immortalization through SV40-T121recovers partially the phenomenon.
Figure 34: the shRNA mediated silencing of endonucleases impairs cells proliferation aspecifically -
#2. Cells treated with shRNAs against resolution nucleases incorporate less EdU than controls. The negative effects on proliferation of shRNA treatments, confirmed by EdU incorporation experiments in primary MEFs (A), could only partially be recovered by SV40T121 immortalization (B).
Nsmce2SD – Mus81KO animals don’t differ significantly from control strains.
Despite the lack of a clear phenotype in MEFs, which suggested the lack of a synergic interaction between Mus81 and Nsmce2, we analyzed the impact of the elimination of MUS81 on the Nsmce2SD background in vivo. We hence focused on the Nsmce2SD-Mus81KO mice and setup ageing curves of double homozygous animals and relevant controls. At
A B
Results - part I the same time, we arranged for heterozygous crosses of double heterozygous mice, in order to verify the transmission of the Mus81KO and the Nsmce2SD alleles. As depicted in figure 35, no remarkable general differences between control animals and Nsmce2SD -Mus81KO double homozygouscould be highlighted. Mice had a similar average weight, and their size wasn’t significantly altered. The coexistence of both mutations, moreover, didn’t affect the animals’ fitness, since they survived as well as controls and bred normally. Finally, the allelic transmission of both mutant genes followed an expected Mendelian pattern, stressing the non-synergic effect of the two mutations, and the lack of synthetic sickness, despite our expectancy.
"
Figure 35: Nsmce2SD: Mus81KO animals show no peculiar differences from control cohorts. Neither their average weight nor size (as in A and B) was affected by the mutated alleles. (C) Animals survival rates where comparable, and the Nsmce2/Mus81 alleles were transmitted at expected Mendelian ratios (D).
The normal phenotype arising from the genetic combination Nsmce2SD - Mus81KO came only partially as a surprise. On one side, our major concern was the lack of major consequences upon the depletion of MUS81, which is at odds with the striking phenotypes reported for the Mus81KO model we employed (McPherson, Lemmers et al.
"70
A
B C
Results - part I 2004). On the other side, though, the absence of a negative synergy between the two alleles further corroborated our observations on the mild phenotype presented by the Nsmce2SD strain.
These results prompted us to explore the involvement of the SMC5/6 complex in dissolution using other mutant strains of SMC5/6.
Smc6S994A as an alternative genetic carrier.
As introduced previously, a major issue to study the SMC5/6 complex in mammals was the lack of appropriate murine working models. Taking in consideration the poor phenotype of Nsmce2SD in vivo, we opted for a recently characterized mutant for the SMC5/6 complex, presenting a stronger phenotype.
The strain bears a point mutation in Smc6 (Smc6S994A-neo) (Ju, Wing et al. 2013) that recapitulates a separation of function mutation (smc6-S1045A) shown to confer increased genomic instability in S. pombe (Jessberger 2002) . It’s to be underlined again how Smc6 is an essential gene, both in yeast and metazoans (Ju et al. 2013) and that the outcomes of the Smc6S994A-neo mutation (an increased rate of SCE at mitosis and sensitivity to replication-challenging agents) phenocopy what observed in the case of several mms21 mutants and functional equivalents (Doe, Murray et al. 1993; Lehmann, Walicka et al.
1995; Verkade, Bugg et al. 1999).
Apart from their genomic instability phenotype , Smc6S994A-neo homozygous animals are remarkably smaller in size, when compared to heterozygous and wt counterparts.
This allelic option represented hence a better experimental scenario for our purposes.
By taking advantage of the Smc6S994A-neo strain, we could employ a mutant for the SMC5/6 showing similar in vitro features as Nsmce2 mutants and possessing, moreover, a more evident and easily characterizable phenotype.
For this reason we rederived the strain on our mixed background and verified the preservation of the size anomaly reported for the model (figure 36).
Results - part I
Figure 36: Smc6S994A-neo animals are smaller than wt counterparts. We rederived the Smc6S994A-neo mutation onto our mixed genetic background and verified the reduced size phenotype. The body size difference between adult (14 weeks) Smc6S994A-neo animals and wt reflected their average weight difference.
According to the strategy outlined for the Nsmce2SD strain, we crossed the new line with Mus81KO animals, and, similarly to the previous case, established matings of heterozygous animals to verify the allelic transmission, as well as ageing curves of double homozygous mice and relevant controls. Strikingly, we encountered a strong synthetic lethality between Smc6S994A-neo and Mus81KO. We weren’t indeed able to obtain neither double homozygous Smc6S994A-neo: Mus81KO animals out of the litters born of our experimental crosses (figure 37) nor double homozygous embryos, at least up to E10.5 (data not shown).
"72
A B
Results - part I
Figure 37: Smc6S994A-neo and Mus81KO are synthetic lethal in mice. The allelic transmission of Mus81KO and Smc6S994A-neo is non Mendelian. The same result was confirmed by the genotyping of embryos up to E10.5 (data not shown).
This striking result represented an obvious major drawback, for the impossibility to obtain animals or cells for experimental purposes. Interestingly though, despite the need for further characterization, the outcome of the Smc6S994A-neo: Mus81KO intercross was supporting the hypothesis that impairing resolution on an "impaired SMC5/6 background" might severely affect fitness.
In order to overcome this issue, we devised to reproduce the same genetic combination by employing a Gen1KO strain we recently obtained (Matthews et al. unpublished) as an alternative strategy to affect resolution.
The Gen1KO mice we employed have no major phenotype, showing a normal lifespan and breeding normally (Matthews et al. unpublished). Our new endeavour was frustrated, though, by the localization of Gen1 and Smc6 in the murine genome. Both genes co-localize on the same arm of chromosome 12 (figure 38) and their genetic linkage hinders their independent segregation, hence hampering the possibility of performing classical genetic analysis. We nevertheless believe that this finding is probably indicative of a functional relationship GEN1 - SMC5/6 complex.
A B
Results - part I
Figure 38: Smc6 and Gen1 show genetic linkage. We checked the localization of Gen1 and Smc6 in the murine genome through the ENSEMBL genome navigator (Flicek, Ahmed et al. 2013) verifying their close proximity, hence the impossibility to study their independent segregation. Intriguingly, this genetic linkage might suggest a potential genetic and functional interplay, as observed in the case of miRNA clusters (Altuvia, Landgraf et al. 2005).
The quest for an alternative genetic carrier – the Nsmce2lox strain.
The remarkable lethal interaction between Smc6S994A-neo and Mus81KO corroborated the hypothesis of a functional bond between the SMC5/6 complex and the metabolism of joint DNA molecules. The lack of homozygous animals to work experimentally, though, represented a major disadvantage. Therefore, in order to proceed, we crossed Mus81KO mice with animals carrying a conditional allele of Nsmce2 that we generated in the lab according to the strategy outlined in figure 39 (Jacome et al. unpublished).
Figure 39: Targeting strategy for the generation of the Nsmce2lox strain. ES cells from wt animals were targeted with a construct bearing the floxed exon3 of Nsmce2 and a FRT-NeoR cassette for clonal selection purposes. Positive recombinants were microinjected in fertilized oocytes to generate the mouse strain
Figure 39: Targeting strategy for the generation of the Nsmce2lox strain. ES cells from wt animals were targeted with a construct bearing the floxed exon3 of Nsmce2 and a FRT-NeoR cassette for clonal selection purposes. Positive recombinants were microinjected in fertilized oocytes to generate the mouse strain